benchmarks.wiki

benchmarks.wiki / Public workspace

DeepSeek V3 / DeepSeek-V3

All models · Read as JSON

These are quotations of published scores, not measurements made here. Evaluation conditions differ by row, so this is not a ranking. Different metrics must not be combined.

These source spellings share a URL slug. Each row retains its reported model name; no alias equivalence is assumed.

Recorded benchmark reports
BenchmarkScoreMetricSetupReported bySource
HumanEvalDeepSeek-V365.2Pass@1Base model; 0-shot.Vendor self-reportReported: Not statedBase Model, Standard Benchmarks, HumanEval (Pass@1), final columnChecked: 2026-09-16
MBPPDeepSeek-V375.4Pass@1Base model; 3-shot.Vendor self-reportReported: Not statedBase Model, Standard Benchmarks, MBPP (Pass@1), final columnChecked: 2026-09-16
CRUXEvalDeepSeek-V369.8Acc.Base model; 2-shot; output prediction.Vendor self-reportReported: Not statedBase Model, Standard Benchmarks, CRUXEval-O (Acc.), final columnChecked: 2026-09-16
CRUXEvalDeepSeek-V367.3Acc.Base model; 2-shot; input prediction.Vendor self-reportReported: Not statedBase Model, Standard Benchmarks, CRUXEval-I (Acc.), final columnChecked: 2026-09-16
AIME 2024DeepSeek V339.2Pass@1Chat model; 8K output limit; multiple runs with varying temperatures for benchmarks under 1000 samples.Vendor self-reportReported: Not statedChat Model, Standard Benchmarks, AIME 2024 (Pass@1), final columnChecked: 2026-09-16
GSM8KDeepSeek-V389.3EMBase model; 8-shot.Vendor self-reportReported: Not statedBase Model, Standard Benchmarks, GSM8K (EM), final columnChecked: 2026-09-16
MATHDeepSeek-V361.6EMBase model; 4-shot.Vendor self-reportReported: Not statedBase Model, Standard Benchmarks, MATH (EM), final columnChecked: 2026-09-16
MGSMDeepSeek-V379.8EMBase model; 8-shot.Vendor self-reportReported: Not statedBase Model, Standard Benchmarks, MGSM (EM), final columnChecked: 2026-09-16
BIG-Bench HardDeepSeek-V387.5EMBase model; 3-shot.Vendor self-reportReported: Not statedBase Model, Standard Benchmarks, BBH (EM), final columnChecked: 2026-09-16
DROPDeepSeek-V389F1Base model; 3-shot.Vendor self-reportReported: Not statedBase Model, Standard Benchmarks, DROP (F1), final columnChecked: 2026-09-16
HellaSwagDeepSeek-V388.9Acc.Base model; 10-shot.Vendor self-reportReported: Not statedBase Model, Standard Benchmarks, HellaSwag (Acc.), final columnChecked: 2026-09-16
MMLU-ProDeepSeek-V364.4Acc.Base model; 5-shot.Vendor self-reportReported: Not statedBase Model, Standard Benchmarks, MMLU-Pro (Acc.), final columnChecked: 2026-09-16
WinoGrandeDeepSeek-V384.9Acc.Base model; 5-shot.Vendor self-reportReported: Not statedBase Model, Standard Benchmarks, WinoGrande (Acc.), final columnChecked: 2026-09-16
MMLUDeepSeek-V387.1Acc.Base model; 5-shot.Vendor self-reportReported: Not statedBase Model, Standard Benchmarks, MMLU (Acc.), final columnChecked: 2026-09-16
ARC-ChallengeDeepSeek-V395.3Acc.Base model; 25-shot.Vendor self-reportReported: Not statedBase Model, Standard Benchmarks, ARC-Challenge (Acc.), final columnChecked: 2026-09-16